ASTNN-Based System for Auditing PHP Code

As PHP language is widely used in web applications, effectively detecting and repairing security vulnerabilities in PHP code has become crucial in research. Thus, we developed an ASTNN-based system for auditing PHP code that transforms input source code into its corresponding Abstract Syntax Tree (AST) to learn and extract features by encoding and decoding at each layer. The system outputs the classification results. We assessed the effectiveness of the proposed model using the test dataset and obtained an accuracy rate of 90%, surpassing other methods based on PHP vulnerability libraries and rules. Moreover, the proposed model exhibits a low false-positive rate and a high recall rate, providing greater advantages than traditional rule-based methods and enhancing the efficiency and precision of vulnerability detection.